Multi-Z Impurity Transport in DIII-D ITER Similar Shape Plasmas: Experiment, Gyrokinetic Simulation, and Gyrokinetic Based, Flux-Matched Profile Predictions

DIII-D experiments probing impurity transport in ITER similar shape H-mode plasmas have been analyzed using a multi-pronged approach: experimental inference of impurity transport, nonlinear gyrokinetic simulation, and gyrokinetic-based, kinetic profile predictions enabled by novel machine learning techniques. A wide range of impurity species (He, Li, C, F, Al, Ca, and W) were introduced into repeat discharges to gather validation quality profile and spectroscopic data and enable experimental inference of impurity transport coefficients (D and V) and impurity peaking. More than 50 nonlinear gyrokinetic simulations were performed with high physics fidelity and compared with experimental heat (Q_i, Q_e) and particle flux (gamma_e), allowing for multi-Z impurity transport coefficient predictions to be extracted. These simulations indicate that ITG and grad-n driven TEM dictate heat and particle transport and reveal a clear dependence of impurity diffusion and peaking on Z that varies radially; ITG dominates inside rho = 0.45 with diffusion scaling with 1/Z, and grad-n TEM dominates at larger radii with diffusion scaling with Z. Simulated impurity peaking trends yield only marginal agreement with measured peaking. Attempts to resolve this discrepancy motivated flux-matched profile prediction based on nonlinear gyrokinetic simulation. Previously prohibitively expensive, such analysis has only recently been made tractable using novel optimization techniques. We present experimental transport results (heat fluxes, particle fluxes, and impurity transport coefficients), nonlinear gyrokinetic simulation comparisons spanning rho = 0.3, 0.4, 0.5, 0.6, 0.7, 0.8, detailed comparison of gyrokinetic flux-matched profiles with experimental kinetic profiles and impurity peaking, and implications for future FPP operation.